1 // SPDX-License-Identifier: GPL-2.0-or-later
3 * fs/cifs/cifs_unicode.c
5 * Copyright (c) International Business Machines Corp., 2000,2009
9 #include <linux/slab.h>
10 #include "cifs_fs_sb.h"
11 #include "cifs_unicode.h"
12 #include "cifs_uniupr.h"
15 #include "cifs_debug.h"
17 int cifs_remap(struct cifs_sb_info *cifs_sb)
21 if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SFM_CHR)
22 map_type = SFM_MAP_UNI_RSVD;
23 else if (cifs_sb->mnt_cifs_flags & CIFS_MOUNT_MAP_SPECIAL_CHR)
24 map_type = SFU_MAP_UNI_RSVD;
26 map_type = NO_MAP_UNI_RSVD;
31 /* Convert character using the SFU - "Services for Unix" remapping range */
33 convert_sfu_char(const __u16 src_char, char *target)
36 * BB: Cannot handle remapping UNI_SLASH until all the calls to
37 * build_path_from_dentry are modified, as they use slash as
65 /* Convert character using the SFM - "Services for Mac" remapping range */
67 convert_sfm_char(const __u16 src_char, char *target)
69 if (src_char >= 0xF001 && src_char <= 0xF01F) {
70 *target = src_char - 0xF000;
109 * cifs_mapchar - convert a host-endian char to proper char in codepage
110 * @target - where converted character should be copied
111 * @src_char - 2 byte host-endian source character
112 * @cp - codepage to which character should be converted
113 * @map_type - How should the 7 NTFS/SMB reserved characters be mapped to UCS2?
115 * This function handles the conversion of a single character. It is the
116 * responsibility of the caller to ensure that the target buffer is large
117 * enough to hold the result of the conversion (at least NLS_MAX_CHARSET_SIZE).
120 cifs_mapchar(char *target, const __u16 *from, const struct nls_table *cp,
128 if ((maptype == SFM_MAP_UNI_RSVD) && convert_sfm_char(src_char, target))
130 else if ((maptype == SFU_MAP_UNI_RSVD) &&
131 convert_sfu_char(src_char, target))
134 /* if character not one of seven in special remap set */
135 len = cp->uni2char(src_char, target, NLS_MAX_CHARSET_SIZE);
142 /* convert SURROGATE_PAIR and IVS */
143 if (strcmp(cp->charset, "utf8"))
145 len = utf16s_to_utf8s(from, 3, UTF16_LITTLE_ENDIAN, target, 6);
157 * cifs_from_utf16 - convert utf16le string to local charset
158 * @to - destination buffer
159 * @from - source buffer
160 * @tolen - destination buffer size (in bytes)
161 * @fromlen - source buffer size (in bytes)
162 * @codepage - codepage to which characters should be converted
163 * @mapchar - should characters be remapped according to the mapchars option?
165 * Convert a little-endian utf16le string (as sent by the server) to a string
166 * in the provided codepage. The tolen and fromlen parameters are to ensure
167 * that the code doesn't walk off of the end of the buffer (which is always
168 * a danger if the alignment of the source buffer is off). The destination
169 * string is always properly null terminated and fits in the destination
170 * buffer. Returns the length of the destination string in bytes (including
173 * Note that some windows versions actually send multiword UTF-16 characters
174 * instead of straight UTF16-2. The linux nls routines however aren't able to
175 * deal with those characters properly. In the event that we get some of
176 * those characters, they won't be translated properly.
179 cifs_from_utf16(char *to, const __le16 *from, int tolen, int fromlen,
180 const struct nls_table *codepage, int map_type)
182 int i, charlen, safelen;
184 int nullsize = nls_nullsize(codepage);
185 int fromwords = fromlen / 2;
186 char tmp[NLS_MAX_CHARSET_SIZE];
187 __u16 ftmp[3]; /* ftmp[3] = 3array x 2bytes = 6bytes UTF-16 */
190 * because the chars can be of varying widths, we need to take care
191 * not to overflow the destination buffer when we get close to the
192 * end of it. Until we get to this offset, we don't need to check
193 * for overflow however.
195 safelen = tolen - (NLS_MAX_CHARSET_SIZE + nullsize);
197 for (i = 0; i < fromwords; i++) {
198 ftmp[0] = get_unaligned_le16(&from[i]);
201 if (i + 1 < fromwords)
202 ftmp[1] = get_unaligned_le16(&from[i + 1]);
205 if (i + 2 < fromwords)
206 ftmp[2] = get_unaligned_le16(&from[i + 2]);
211 * check to see if converting this character might make the
212 * conversion bleed into the null terminator
214 if (outlen >= safelen) {
215 charlen = cifs_mapchar(tmp, ftmp, codepage, map_type);
216 if ((outlen + charlen) > (tolen - nullsize))
220 /* put converted char into 'to' buffer */
221 charlen = cifs_mapchar(&to[outlen], ftmp, codepage, map_type);
224 /* charlen (=bytes of UTF-8 for 1 character)
225 * 4bytes UTF-8(surrogate pair) is charlen=4
226 * (4bytes UTF-16 code)
227 * 7-8bytes UTF-8(IVS) is charlen=3+4 or 4+4
228 * (2 UTF-8 pairs divided to 2 UTF-16 pairs) */
231 else if (charlen >= 5)
236 /* properly null-terminate string */
237 for (i = 0; i < nullsize; i++)
244 * NAME: cifs_strtoUTF16()
246 * FUNCTION: Convert character string to unicode string
250 cifs_strtoUTF16(__le16 *to, const char *from, int len,
251 const struct nls_table *codepage)
255 wchar_t wchar_to; /* needed to quiet sparse */
257 /* special case for utf8 to handle no plane0 chars */
258 if (!strcmp(codepage->charset, "utf8")) {
260 * convert utf8 -> utf16, we assume we have enough space
261 * as caller should have assumed conversion does not overflow
262 * in destination len is length in wchar_t units (16bits)
264 i = utf8s_to_utf16s(from, len, UTF16_LITTLE_ENDIAN,
265 (wchar_t *) to, len);
267 /* if success terminate and exit */
271 * if fails fall back to UCS encoding as this
272 * function should not return negative values
273 * currently can fail only if source contains
274 * invalid encoded characters
278 for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
279 charlen = codepage->char2uni(from, len, &wchar_to);
281 cifs_dbg(VFS, "strtoUTF16: char2uni of 0x%x returned %d\n",
283 /* A question mark */
287 put_unaligned_le16(wchar_to, &to[i]);
291 put_unaligned_le16(0, &to[i]);
296 * cifs_utf16_bytes - how long will a string be after conversion?
297 * @utf16 - pointer to input string
298 * @maxbytes - don't go past this many bytes of input string
299 * @codepage - destination codepage
301 * Walk a utf16le string and return the number of bytes that the string will
302 * be after being converted to the given charset, not including any null
303 * termination required. Don't walk past maxbytes in the source buffer.
306 cifs_utf16_bytes(const __le16 *from, int maxbytes,
307 const struct nls_table *codepage)
310 int charlen, outlen = 0;
311 int maxwords = maxbytes / 2;
312 char tmp[NLS_MAX_CHARSET_SIZE];
315 for (i = 0; i < maxwords; i++) {
316 ftmp[0] = get_unaligned_le16(&from[i]);
319 if (i + 1 < maxwords)
320 ftmp[1] = get_unaligned_le16(&from[i + 1]);
323 if (i + 2 < maxwords)
324 ftmp[2] = get_unaligned_le16(&from[i + 2]);
328 charlen = cifs_mapchar(tmp, ftmp, codepage, NO_MAP_UNI_RSVD);
336 * cifs_strndup_from_utf16 - copy a string from wire format to the local
338 * @src - source string
339 * @maxlen - don't walk past this many bytes in the source string
340 * @is_unicode - is this a unicode string?
341 * @codepage - destination codepage
343 * Take a string given by the server, convert it to the local codepage and
344 * put it in a new buffer. Returns a pointer to the new string or NULL on
348 cifs_strndup_from_utf16(const char *src, const int maxlen,
349 const bool is_unicode, const struct nls_table *codepage)
355 len = cifs_utf16_bytes((__le16 *) src, maxlen, codepage);
356 len += nls_nullsize(codepage);
357 dst = kmalloc(len, GFP_KERNEL);
360 cifs_from_utf16(dst, (__le16 *) src, len, maxlen, codepage,
363 len = strnlen(src, maxlen);
365 dst = kmalloc(len, GFP_KERNEL);
368 strlcpy(dst, src, len);
374 static __le16 convert_to_sfu_char(char src_char)
380 dest_char = cpu_to_le16(UNI_COLON);
383 dest_char = cpu_to_le16(UNI_ASTERISK);
386 dest_char = cpu_to_le16(UNI_QUESTION);
389 dest_char = cpu_to_le16(UNI_LESSTHAN);
392 dest_char = cpu_to_le16(UNI_GRTRTHAN);
395 dest_char = cpu_to_le16(UNI_PIPE);
404 static __le16 convert_to_sfm_char(char src_char, bool end_of_string)
408 if (src_char >= 0x01 && src_char <= 0x1F) {
409 dest_char = cpu_to_le16(src_char + 0xF000);
414 dest_char = cpu_to_le16(SFM_COLON);
417 dest_char = cpu_to_le16(SFM_DOUBLEQUOTE);
420 dest_char = cpu_to_le16(SFM_ASTERISK);
423 dest_char = cpu_to_le16(SFM_QUESTION);
426 dest_char = cpu_to_le16(SFM_LESSTHAN);
429 dest_char = cpu_to_le16(SFM_GRTRTHAN);
432 dest_char = cpu_to_le16(SFM_PIPE);
436 dest_char = cpu_to_le16(SFM_PERIOD);
442 dest_char = cpu_to_le16(SFM_SPACE);
454 * Convert 16 bit Unicode pathname to wire format from string in current code
455 * page. Conversion may involve remapping up the six characters that are
456 * only legal in POSIX-like OS (if they are present in the string). Path
457 * names are little endian 16 bit Unicode on the wire
460 cifsConvertToUTF16(__le16 *target, const char *source, int srclen,
461 const struct nls_table *cp, int map_chars)
468 wchar_t *wchar_to; /* UTF-16 */
472 if (map_chars == NO_MAP_UNI_RSVD)
473 return cifs_strtoUTF16(target, source, PATH_MAX, cp);
475 wchar_to = kzalloc(6, GFP_KERNEL);
477 for (i = 0; i < srclen; j++) {
478 src_char = source[i];
481 /* check if end of string */
485 /* see if we must remap this char */
486 if (map_chars == SFU_MAP_UNI_RSVD)
487 dst_char = convert_to_sfu_char(src_char);
488 else if (map_chars == SFM_MAP_UNI_RSVD) {
492 * Remap spaces and periods found at the end of every
493 * component of the path. The special cases of '.' and
494 * '..' do not need to be dealt with explicitly because
495 * they are addressed in namei.c:link_path_walk().
497 if ((i == srclen - 1) || (source[i+1] == '\\'))
498 end_of_string = true;
500 end_of_string = false;
502 dst_char = convert_to_sfm_char(src_char, end_of_string);
506 * FIXME: We can not handle remapping backslash (UNI_SLASH)
507 * until all the calls to build_path_from_dentry are modified,
508 * as they use backslash as separator.
511 charlen = cp->char2uni(source + i, srclen - i, &tmp);
512 dst_char = cpu_to_le16(tmp);
515 * if no match, use question mark, which at least in
516 * some cases serves as wild card
521 /* convert SURROGATE_PAIR */
522 if (strcmp(cp->charset, "utf8") || !wchar_to)
524 if (*(source + i) & 0x80) {
525 charlen = utf8_to_utf32(source + i, 6, &u);
530 ret = utf8s_to_utf16s(source + i, charlen,
537 dst_char = cpu_to_le16(*wchar_to);
539 /* 1-3bytes UTF-8 to 2bytes UTF-16 */
540 put_unaligned(dst_char, &target[j]);
541 else if (charlen == 4) {
542 /* 4bytes UTF-8(surrogate pair) to 4bytes UTF-16
543 * 7-8bytes UTF-8(IVS) divided to 2 UTF-16
544 * (charlen=3+4 or 4+4) */
545 put_unaligned(dst_char, &target[j]);
546 dst_char = cpu_to_le16(*(wchar_to + 1));
548 put_unaligned(dst_char, &target[j]);
549 } else if (charlen >= 5) {
550 /* 5-6bytes UTF-8 to 6bytes UTF-16 */
551 put_unaligned(dst_char, &target[j]);
552 dst_char = cpu_to_le16(*(wchar_to + 1));
554 put_unaligned(dst_char, &target[j]);
555 dst_char = cpu_to_le16(*(wchar_to + 2));
557 put_unaligned(dst_char, &target[j]);
562 dst_char = cpu_to_le16(0x003f);
568 * character may take more than one byte in the source string,
569 * but will take exactly two bytes in the target string
572 put_unaligned(dst_char, &target[j]);
576 put_unaligned(0, &target[j]); /* Null terminate target unicode string */
582 * cifs_local_to_utf16_bytes - how long will a string be after conversion?
583 * @from - pointer to input string
584 * @maxbytes - don't go past this many bytes of input string
585 * @codepage - source codepage
587 * Walk a string and return the number of bytes that the string will
588 * be after being converted to the given charset, not including any null
589 * termination required. Don't walk past maxbytes in the source buffer.
593 cifs_local_to_utf16_bytes(const char *from, int len,
594 const struct nls_table *codepage)
600 for (i = 0; len && *from; i++, from += charlen, len -= charlen) {
601 charlen = codepage->char2uni(from, len, &wchar_to);
602 /* Failed conversion defaults to a question mark */
606 return 2 * i; /* UTF16 characters are two bytes */
610 * cifs_strndup_to_utf16 - copy a string to wire format from the local codepage
611 * @src - source string
612 * @maxlen - don't walk past this many bytes in the source string
613 * @utf16_len - the length of the allocated string in bytes (including null)
614 * @cp - source codepage
615 * @remap - map special chars
617 * Take a string convert it from the local codepage to UTF16 and
618 * put it in a new buffer. Returns a pointer to the new string or NULL on
622 cifs_strndup_to_utf16(const char *src, const int maxlen, int *utf16_len,
623 const struct nls_table *cp, int remap)
628 len = cifs_local_to_utf16_bytes(src, maxlen, cp);
630 dst = kmalloc(len, GFP_KERNEL);
635 cifsConvertToUTF16(dst, src, strlen(src), cp, remap);